What is chaiNNer?
chaiNNer is a desktop app for processing images by connecting nodes in a visual flowchart. It started as an AI upscaling tool and grew into a general image pipeline builder: load an image, upscale it with an AI model, adjust it and save it, all as linked steps.
It supports PyTorch, NCNN, ONNX and TensorRT models, so you can use the many community upscaling and restoration models. Chains can process whole folders of images or video frames in one run. It works with Nvidia, AMD, Apple silicon and Intel GPUs, with a CPU fallback, and a built-in dependency manager installs what it needs.
Main Features
- Visual node editor for image pipelines
- AI upscaling with community models
- Batch processing of folders and video
- PyTorch, NCNN, ONNX and TensorRT support
- Works on most GPUs with CPU fallback
- Built-in dependency manager
How to Get Started with chaiNNer
- Download chaiNNer for your system from chainner.app.
- Open the dependency manager and install the packages for your model type.
- Drag in Load Image, an upscale model node and Save Image, and connect them.
- Run the chain on one image or a whole folder.
chaiNNer Pricing
chaiNNer is free and open source under the GPL-3.0 licence. The developers accept optional donations.
Who Should Use It?
- People upscaling old photos or game textures
- Anime and art fans restoring images
- Video hobbyists processing frames
- Power users building repeatable image workflows
- Anyone who wants free local upscaling
chaiNNer Alternatives
- Topaz Labs: Paid desktop AI upscaling and enhancement.
- Magnific AI: AI upscaler that adds detail.
- Squoosh: Free in-browser image compressor and resizer.
Frequently Asked Questions
Is chaiNNer free?
Yes. chaiNNer is free and open source under GPL-3.0. Donations are optional.
Does chaiNNer include upscaling models?
You download models separately, for example from community model collections, then load them in a node.
Which systems does it run on?
Windows, macOS and Linux, on Nvidia, AMD, Apple silicon or Intel GPUs, or on the CPU.